Time series data is a series of time-ordered data collected at different points in time. Such data reflects the change of an object, phenomenon, indicator and so on over time. Because time series data generally is data sampled at equal time intervals, existing methods for storing time series data are generally: reserving a storage area for each time point according to the above-mentioned time interval; upon receiving time series data of a time point, writing the data to the storage area reserved for the time point.
However, currently in practice, time series data may not always be generated after every time interval, or even the time interval may change. Consequently, there are a large amount of data including null values in the storage area. If such data including null values is directly written to a magnetic disk, large storage space is occupied, and I/O traffic during subsequent data read/write is increased.